让机器人模型在低精度下仍保持高精度，提升效率。
Saliency-Aware Quantized Imitation Learning for Efficient Robotic Control
- 根据重要状态自动加权损失，重点训练关键决策
- 4位量化后速度提升2.5倍，能耗降低2.5倍，精度几乎不变
- 适合边缘设备部署，尤其适用于机器人控制
基于深度神经网络（DNN）的策略模型，如视觉-语言-动作（VLA）模型，在多模态输入下表现出色，但规模扩大导致计算开销剧增，难以在机器人操作等资源受限场景中部署。为此，我们提出显著性感知量化模仿学习（SQIL），结合量化感知训练与针对关键状态的损失加权策略。通过显著性分数识别关键状态，并在训练损失中强化其权重，使模型在低比特精度下仍保持决策准确性。我们在多种仿真基准、真实任务及跨领域任务（自动驾驶、物理模拟）中验证了其泛化能力，始终能恢复全精度性能。值得注意的是，4位权重量化后的VLA模型在边缘GPU上实现最高2.5倍加速和2.5倍能耗降低，精度损失极小。这表明SQIL在资源受限设备上高效部署大规模模仿学习策略模型方面具有巨大潜力。
原文摘要 · Abstract (English)
Deep neural network (DNN)-based policy models, such as vision-language-action (VLA) models, excel at automating complex decision-making from multi-modal inputs. However, scaling these models greatly increases computational overhead, complicating deployment in resource-constrained settings like robot manipulation and autonomous driving. To address this, we propose Saliency-Aware Quantized Imitation Learning (SQIL), which combines quantization-aware training with a selective loss-weighting strategy for mission-critical states. By identifying these states via saliency scores and emphasizing them in the training loss, SQIL preserves decision fidelity under low-bit precision. We validate SQIL's generalization capability across extensive simulation benchmarks with environment variations, real-world tasks, and cross-domain tasks (self-driving, physics simulation), consistently recovering full-precision performance. Notably, a 4-bit weight-quantized VLA model for robotic manipulation achieves up to 2.5x speedup and 2.5x energy savings on an edge GPU with minimal accuracy loss. These results underline SQIL's potential for efficiently deploying large IL-based policy models on resource-limited devices.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。